Active Agent

Appearance

Atlas Cloud Provider ​

The Atlas Cloud provider connects Active Agent to Atlas Cloud's OpenAI-compatible Chat Completions API. It supports the standard Active Agent chat features, including streaming, structured output, and tool calling when the selected model supports them.

Installation ​

Atlas Cloud uses the openai gem:

bash
bundle add openai

Configuration ​

Set the API key in your environment:

bash
ATLASCLOUD_API_KEY=your-api-key

Configure the provider in config/active_agent.yml:

yaml
development:
  atlas_cloud:
    service: "AtlasCloud"
    api_key: <%= ENV["ATLASCLOUD_API_KEY"] %>
    model: "qwen/qwen3.8-max"

Then select it from an agent:

ruby
class AtlasCloudAgent < ApplicationAgent
  generate_with :atlas_cloud, model: "qwen/qwen3.8-max"

  def ask
    prompt(message: params[:message])
  end
end

The provider defaults to https://api.atlascloud.ai/v1. You can override base_url in the provider configuration (for example base_url: "https://proxy.example.com/v1") when routing through a compatible proxy.

Chat Only ​

The provider supports chat completions only. Atlas Cloud does not serve an embeddings endpoint (/v1/embeddings returns 404), so embed is not available through this provider.

Model Selection ​

Atlas Cloud model identifiers use a provider/model format. Query the current catalog before selecting a model:

bash
curl https://api.atlascloud.ai/v1/models \
  -H "Authorization: Bearer $ATLASCLOUD_API_KEY"

Use the returned model ID as the model value. Model capabilities vary, so verify streaming, structured output, or tool support for the chosen model.